Genetic neuro-fuzzy approach for unmanned fixed wing attitude control
With the imminent growth and the progressive interest in the subject, the unmanned aerial vehicles (UAVs) are already a reality in our daily life. The search for air vehicles, which is ambitious for a future in which the UAVs can act autonomously and safely, continuously drives this sector. The pres...
Saved in:
Published in | 2017 International Conference on Military Technologies (ICMT) pp. 485 - 492 |
---|---|
Main Authors | , |
Format | Conference Proceeding |
Language | English |
Published |
IEEE
01.05.2017
|
Subjects | |
Online Access | Get full text |
DOI | 10.1109/MILTECHS.2017.7988808 |
Cover
Loading…
Summary: | With the imminent growth and the progressive interest in the subject, the unmanned aerial vehicles (UAVs) are already a reality in our daily life. The search for air vehicles, which is ambitious for a future in which the UAVs can act autonomously and safely, continuously drives this sector. The present work aims to apply artificial intelligence techniques to classical control systems, in order to control the attitude of a fixed wing aircraft adaptively. The open source software ArduPlane was used as the basis for this project, which has an enhanced implementation of a PID controller for attitude. It is noteworthy the need to frequent adjustment for gains tied to the attitude control system; either for basic adjustment constants, as well as for parameters whose calibration needs specific technical knowledge of the system. In order to automate this process and ensure the optimization of these parameters throughout the mission, a genetic neuro-fuzzy approach was proposed to make this procedure implicit and transparent to the flight operator. With the use of an Adaptive Neuro-Fuzzy Inference System (ANFIS) capable of predicting the attitude of the aircraft, together with an optimization system (genetic algorithm), it is possible to construct an efficient control architecture, which ensures an improved control that always searches for optimized parameters throughout the mission autonomously. |
---|---|
DOI: | 10.1109/MILTECHS.2017.7988808 |